One of the most interesting demos at the TDK booth is TDK Qeexo AutoML, an end-to-end machine learning (ML) platform that allows customers to build production-level ML / Artificial Intelligence (AI) applications while spending minimal engineering resources. The platform is completely automated and seems very easy to use and operate.
The general idea is to enable TDK sensor customers to build specific AI applications themselves. TDK components are used in an extremely high number of applications, and in practice, it would be impossible for TDK to build specific AI applications for each use case. Using this platform, TDK can handle the heavy lifting of operating an ML workload while leaving specific workflow details to their customers.
Qeexo AutoML has a graphical user interface that takes users through multiple setup screens where they can define where data is coming from, most likely from one or several TDK sensors. Data signals can be visualized as charts for analysis or debugging. Data samples can then be used to train ML models to recognize various sensor data stream combinations.
At CES, the Qeexo team had a demo featuring pumps and pipes that could represent a typical Industrial environment. Using their platform, they demonstrated how easy it was for the application to detect that things were running or if there were any mechanical issues by looking at the water flow, vibrations, etc. In theory, monitoring things such as “motor health” and much more is also possible.
This platform looks very promising, and it’s a great move by TDK to enable their customers to build all kinds of applications that may significantly boost productivity and reduce maintenance in the long run. Additionally, this could spark more customers to buy equipment since they can exploit data immediately and efficiently.